Compression Algorithm for Multi-view Video Coding Using a Depth Map

In this paper, we propose an efficient multi-view video coding algorithm that uses a depth map. This algorithm is based on the correlation between texture maps and depth maps. This correlation is based on the fact that the motion vectors (MVs) in texture maps are strongly affected by their depth values. In general, background objects have smaller MVs whereas foreground objects have higher MVs. Background objects also have smaller depth values whereas foreground objects have larger depth values. We demonstrate this dependency relationship experimentally. Based on this correlation, we developed an efficient motion prediction method. Thus, we propose a compression algorithm for multi-view video coding based on this motion prediction algorithm.